Caching mongoose objects with Redis

As Mongoose objects wrap a MongoDB document, there'd be no reason that you couldn't call

JSON.stringify(mongooseObject.toJSON())

which would return a string representing the MongoDB document. (toJSON) You could then store that result in a key/value in redis.

Where it starts to get more complex is that you'd need to first override the normal save and update functionality to save any modifications to your redis store rather than to the database. While doable, Mongoose wasn't designed for that and you'd be probably more successful to just use the native mongodb drivers and manage general document interactions that way. There are a number of extremely handy operators that you'd need to independently handle (like $push for example, which would add a single value to an array).

The real issue though is that you loose nearly all of the power of MongoDB by not being able to use the query engine or aggregation framework if all of the data isn't already stored in MongoDB (and even if it is, you're still bypassing your caching layer). And, if you're not using any of that functionality, then MongoDB may not be the best match for your needs (and you might instead consider something like CouchDB).

While I can see the potential value of using a caching layer for a high performance MongoDB system, the complexity of a write-back style cache may be more complex than it's worth (and not necessarily safe).

Of course, a write-through cache would be simpler (although you have the complexity of two data-stores and making sure writes are committed consistently if you're going to trust the cache and DB).

(As an aside, I'm not sure how you'd actually manage timeouts, as I thought redis deleted the values associated with keys if they were assigned a lifetime/timeout? I wouldn't want to loose data to the redis cache if you were doing write-back).